Transimpedance amplifier design

[mmzaj]

hi everybody

i have a term project , where i am supposed to design an optical communication system of 10Mbps , a BER of 10^-9 , a sensitivity smaller than -15dBm . i have designed an optical Rx with a transimpedance amplifier , post amplifier ( differentiator in this case , to compensate for the integrator effect of the TIA ) , and limiting amplifier . i would like to have your opinion on this design .

B.R

 

[berkeman]

hi everybody

i have a term project , where i am supposed to design an optical communication system of 10Mbps , a BER of 10^-9 , a sensitivity smaller than -15dBm . i have designed an optical Rx with a transimpedance amplifier , post amplifier ( differentiator in this case , to compensate for the integrator effect of the TIA ) , and limiting amplifier . i would like to have your opinion on this design .

B.R

Sorry, for me, the image is not clear. But still, a PIN photodiode amp cannot be differential optimally, right? Are you referring to subsequent stages?

And your BER will depend on your attenuation in the fibre and the length, and the noise of the RX amp. Can you please post about that quantitavly?

 

[mmzaj]

Sorry, for me, the image is not clear. But still, a PIN photodiode amp cannot be differential optimally, right? Are you referring to subsequent stages?

And your BER will depend on your attenuation in the fibre and the length, and the noise of the RX amp. Can you please post about that quantitavly?

there is no differential amp in the design , neither the PIN amp ( TIA ) nor other stages . the TIA has to be a low noise amp , therefore the working resistor is in the order of 150K , to eliminate johnson noise , also high sensitivity requires a huge transimpedance gain , which is satisfied by a high working resistor . the problem here is that the high resistor ruins the flat bandwidth of the amp , and TIA acts as an integrator with a 6dB/oct slope of low pass . the differentiator amp compensates for that effect , and restores a flat characteristic . finally a limiting amp is placed at the final stage to limit the swing of the output to ~ 2Vp-p .
as for the BER , a rough estimation of the SNR is made by minimizing the noise at the Rx front end .